Buildfarm build_test script on an embedded device.

Christopher R. Hertel crh at
Sun Oct 2 10:06:50 MDT 2011


The device is a Seagate Dockstar.  It has 128MB of RAM soldered onto the
mainboard, and I have a 2GB swap partition allocated on the USB-attached
2.5" SATA drive (less than 512M of which is in use).  Yes, that'll be slow.
 I expect that.  I just don't know *how* slow since I have no experience
with the buildfarm tests.

What I am really looking for is some sense of how far along the tests have
gotten, and how much more there is to be done.  A status of some sort.

If I knew the tests fairly well, I would be able to judge based on what I
see in 'top'.

Chris -)-----

Dan Shearer wrote:
> On Sun, Oct 02, 2011 at 10:20:25AM -0500, Christopher R. Hertel wrote:
>> I thought it would be interesting to add an ARM-based system to the build
>> farm.  I've talked with some of you offline about doing this.
>    :
>> It's still running.  It has been well over 8 hours and it's still running.
> What is your ARM device?
> A different question that might be relevant is "how can I make an ARM
> target run fast", which is (more or less) a topic discussed recently on
> the QEMU dev list[1]. This refers to the 'system' emulation mode of
> QEMU.  One paper referred to calculates that an x86 host of 2GHz will be
> faster than a real ARM board (lots of if's and but's here). I would add
> to this that depending where the bottlenecks are you can perform some
> tricks such as running disk out of an undelying ramdisk, increasing RAM
> beyond what an ARM target would normally have and using passthrough drivers.
> Intensive CPU loads will run slowly on QEMU, and floating point worse.
> --
> Dan Shearer
> dan at
> [1]

"Implementing CIFS - the Common Internet FileSystem" ISBN: 013047116X
Samba Team --     -)-----   Christopher R. Hertel
jCIFS Team --   -)-----   ubiqx development, uninq.
ubiqx Team --     -)-----   crh at
OnLineBook --    -)-----   crh at

More information about the samba-technical mailing list